[GLD Serializer Benchmark] C++ nlohmann_json 3.12.0 — spec compliance misses

Hi — this is a friendly report from the GLD Serializer Benchmark, an independent multi-language suite that times serializers and checks them against the published format specs (RFC 8259, YAML 1.2, …).

We measured nlohmann_json 3.12.0 on C++. The suite recorded spec-profile misses for this pin. The catalog is report-only: a miss is not a CI failure, and some cases are older or stricter profiles (for example RFC 4627 vs RFC 8259) rather than a crash.

Compliance results (please start here)

This link opens the live heatmap on C++ with nlohmann_json already selected (All standards). Click a scored cell for each FAIL and the cited spec paragraph:

https://leo-gan.github.io/GLD.SerializerBenchmark/dashboard/#compliance/cpp/nlohmann_json

How the catalog is built, and how to read a FAIL line:

• https://leo-gan.github.io/GLD.SerializerBenchmark/compliance/
• https://leo-gan.github.io/GLD.SerializerBenchmark/compliance/reading-results/

I am not pasting the case list here so you can use the Dashboard (filters, heatmap, and spec links) rather than a static dump.

Optional: one standard family only — #compliance/cpp/nlohmann_json/json (or yaml, cbor, …).

Benchmark timings (same pin)

The same suite also publishes encode / decode time and payload size for this library on C++:

https://leo-gan.github.io/GLD.SerializerBenchmark/dashboard/?lang=cpp&data=document@n=1&mode=bytes#dashboard

Details table (this library pre-selected):

https://leo-gan.github.io/GLD.SerializerBenchmark/dashboard/?lang=cpp&ser=nlohmann_json&data=document@n=1&mode=bytes#detailed-analytics

That page is the live L1 snapshot, not a claim that one library is “best.”

Ask

Please look at the Compliance view for nlohmann_json. If a miss is a real decoder/encoder bug you want to fix, great. If it is an intentional default (accepting NaN, YAML 1.1 bools, …), a short note here is enough — we will keep the catalog as-is unless the library’s documented contract changes.

Thanks for the library.

Start with the C++ nlohmann_json compliance heatmap and open each scored FAIL cell to read its case and cited specification paragraph. Use the compliance and reading-results pages to determine whether each miss reflects a real behavior issue or an intentional default. Done means the reported misses are classified, with any confirmed bug addressed and any intentional behavior explained in the issue.